Minutes interim-2017-dots-02: Thu 10:00

Meeting Minutes DDoS Open Threat Signaling (dots) WG
Title Minutes interim-2017-dots-02: Thu 10:00
State Active
Other versions plain text
Last updated 2017-06-08

Meeting Minutes

   DDoS Open Threat Signaling (DOTS) WG Virtual Interim Meeting
Thursday, June 8, 2017
14:00 - 15:30 UTC

==[ Introduction ]==
** Presenter: Roman Danyliw and Tobias Gondrom
** Presentation:

The chairs presented a summary of the working group's activities and discussed
preparation for IETF 99.

Approximately 18 participants were online through-out the virtual interim

==[ Use Case Discussion ]==
** Presenter: Roland Dobbins
** Presentation:
** Draft: draft-ietf-dots-use-cases-05

Dobbins summarized changes in the -05 draft.  Version -06 is anticipated June
16th, and a -07 June 23rd.  The -07 will likely be sent out for WGLC.

==[ Requirements Discussion ]==
** Presenter: Andrew Mortensen
** Presentation:
** Draft: draft-ietf-dots-requirements-05

Mortensen discussed the changes in the -05 draft and recent feedback on the
mailing list.

Q: (Tobias Gondrom): When will it be appropriate to start WGLC?
A: (Andrew Mortensen): We'll have a better sense after the -06 is published
(before IETF 99).

Q: (Tiru Reddy): Will the new homenet use cases introduce any new requirements?
A: (Roland Dobbins): No, beyond changing the scale of the number of DOTS
clients. A: (Andrew Mortensen): Agreed.  This use case aligns with the new
requirements on handling multiple overlapping requests. A: (Roland Dobbins):
Handled by the orchestration system, there will be a need for the DOTS server
to do application level rate limiting.

Q: (Frank Xia): Per issue #5 (distinction between session and channel), how was
this addressed? A: (Andrew Mortensen): After analysis, most references to
session in the requirement were identified to be referencing the signal
channel.  Do you feel we need improved definitions? A: (Frank Xia): Yes, given
the use of this language in the signal channel draft.

==[ Architecture Discussion ]==
** Presenter: Andrew Mortensen
** Presentation:
** Draft: draft-ietf-dots-architecture-03

Mortensen discussed the changes in the -03 draft.

Q: (Andrew Mortensen): Are there changes to make in the architecture due to
multi-homing? A: (Roland Dobbins): Multi-homing doesn't matter in DOTS.  We're
conflating end-point ID with routing information. A: (Mohamed Boucadair): There
will be a need to address this architecture. A: (Roland Dobbins): This would
require a lot more state to be kept. A: (Tiru Reddy): We need to have a
multi-homing discussion.  This design will strongly influence client and
gateway implementations A: (Roland Dobbins): Agreed.  It may also be a
configuration issue. A: (Tiru Reddy): We need an easy way to do configuration.
A: (Roland Dobbins): This might be implementation guidance.  We need to
determine where to put this text. A: (Roman Danyliw): Let's continue this
conversation on the mailing list; it will also be a topic for a design team
meeting at IETF 99; and a formal agenda item on this topic can be added to the
IETF 99 meeting. A: (Nik Teague): Can someone better frame this problem for
discussion? A: (Roman Danyliw): Call for volunteers to post to the mailing list
--> Mohamed Boucadair and Tiru Reddy volunteer.

Comment (Mohamed Boucadair): An improved discussion of privacy is needed by
this draft.

==[ Protocol: Signal and Data Channel ]==
** Presenter: Tirumal Reddy
** Presentation:
** Draft: draft-ietf-dots-data-channel-01

Reddy presented the changes to these drafts since they were adopted as WG items.

Q: (Andrew Mortensen): Does NTT Communications have an update don their
implementation? A: (Kaname Nishizuka): NTT is continuing development.  A
challenge with the data channel implementation is finding a good RESTCONF

Q: (Roman Danyliw): Are there any new implementations of this protocol?
A: (Andrew Mortensen): Arbor has a basic signal channel implementation.

==[ Protocol: Service Discovery ]==
** Presenter: Tirumal Reddy
** Presentation:

Reddy presented a candidate approach to DOTS service discovery.

Q: (Tiru Reddy): What's the feedback on this approach as something better than
local configuration? A: (Roland Dobbins): Anycast might be challenging in an
inter-domain situation. A: (Roland Dobbins): Something like this would be
needed if DOTS is added to 1000s of CPEs of an ISP.  This approach doesn't
change the architecture or the base protocol. A: (Tiru Reddy): Agreed. A:
(Mohamed Boucadair): Also agree that this is important, and doesn't change the
architecture.  It's likely appropriate in a seperate draft.  There could be
multiple approaches.  DHCP, per draft-boucadair-dots-dhcp-01, could also be
another option. A: (Roland Dobbins): Agreed. A: (Roman Danyliw): There appears
to be WG interest.  What's the next step? A: (Andrew Mortensen): Perhaps a
simple draft captures the options? A: (Roman Danyliw): Volunteers?  -->
Mohamed Boucadair and Tiru Reddy will discuss how to bring this to the WG. A:
(Roman Danyliw): We can also make this another design team meeting at IETF 99.

==[ Closing ]==
The chairs summarized the actions from the meeting:

** The use cases, requirements and achitecture drafts appear to be maturing
making a WGLC possible in the late summer ** Updates to these drafts will be
made prior to the 7/3/2017 draft cut-off for IETF 99 ** Design team meetings
around multi-homing and service discovery will be schedule during the week of